Selected Topics of Artificial Intelligence

Organisation

Announcements

• First lesson in term WS 19//20: Monday, 14.10. 2019

Grading/Exam:

After the presentations, which are listed in the table below, student-groups can select one of the presented topics. The selected topic is studied in more detail in a homework project. The results of this study are finally presented to all students. This homework and its presentation is assessed.

Idea of this lecture

Artificial Intelligence (AI) provides tools, which can be applied in a vast area of application. Traditional application areas are e.g. robotics, autonomous driving, speech recognition, object recognition, chatbots, information extraction, data mining, recommender systems …. However, in the age of digitization nearly all fields of computer science benefit from smart AI techniques.

The goal of this lecture is to demonstrate how AI can be applied in different fields of computer science and media. Each lecture is presented by an expert in the corresponding application area.

Timetable

All lectures will be held at Monday, 11.45h-13.15h. The concrete date for each topic will be defined soon. At the end of the term, students have to present their homework.

Short Descriptions of individual lectures

Prof. Dr. Roland Schmitz: Applications of AI in Security

According to Bruce Schneier, “Artificial intelligence technologies have the potential to upend the longstanding advantage that attack has over defense on the Internet. This has to do with the relative strengths and weaknesses of people and computers, how those all interplay in Internet security, and where AI technologies might change things.” (IEEE Security & Privacy, March 2018).

In my lecture, I will provide a brief overview of some of today’s applications of AI in IT-Security, namely: * AI for detecting attack traffic in networks * AI for automatic detection of software vulnerabilities * AI for automatic detection of spearphishing attempts in enterprise networks

Currently, using AI in IT-Security is often based on the idea of viewing an attack as an anomaly among benign traffic or software. Naturally, this approach has a certain false positive rate, i.e. a certain probability that benign data are wrongly classified as attacks and false alarms are triggered. While in some scenarios a false positive rate of 1% may be tolerable, the last scenario (about 100 Million archived e-mails in an enterprise network) calls for alternative approaches.

Prof. Dr. Gottfried Zimmermann: AI’s role in Assistive Technologies

Short description: Assistive Technology (AT) has a long tradition in providing support for people with disabilities and older people, for example magnifier glasses and walking aids. More recently, electronic devices and special software have aided in providing access to computers and IT-enabled devices, for example head pointing devices and speech recognition software. AI plays an increasingly important role in AT in two ways: First, it makes AT better and easier to use. Thanks to AI, speech recognition has gained recognition rates that were unseen before. Second, it can help the user to find and set up a configuration of AT that is best for the user in a specific situation; a task that can be daunting given the complexity of AT and the variety of hardware and software options. In this lecture, we will look at current examples of AT that involves AI, and we will hypothesize how the future of AT may be further affected by AI.

Dr. Andreas Stiegler: Procedural Design

Art and Design are hard to define fields, perhaps due to their creative and aesthetic focus. Thus, it might seem surprising to see them as an AI problem, in particular as “artificial creativity” is still an open topic. However, there are countless applications! In this little overview, we’ll cluster the problem domain into three bullets: Procedural Design (quite similar to Procedural Content Generation in games: how to design together with an AI), Artificial Design (what’s the state of AIs doing the creative task for you?), and AIs in the creative workflow (how can AI-based tools be used by artists when coming up with designs). We’ll take a look at some examples - in particular for Procedural Design - and draw links to the algorithms used, as well as other areas you might find these approaches in.

Prof. Dr. Oliver Kretzschmar: Machine Learning in IT Product-Management

This lecture gives a short overview, which traditional as well as newer Machine Learning methods can be used in the context of a new product development, the product optimization as well as with the product marketing. Exemplarily some examples with practice context are indicated and discussed. Possibilities as well as limits will be pointed out in particular in the creative context of so-called disruptive as well as incremental innovations.

Prof. Dr. Sabiha Ghellal: AI for Data Visualisation / Visual Analytics

Visual representations and interaction techniques take advantage of the human eye’s broad bandwidth pathway into the mind to allow users to see, explore, and understand large amounts of information at once. (Thomas & Cook, 2005). Visualization, or visual data analysis, is the most reliant on the cognitive skills of human analysts, and allows the discovery of unstructured actionable insights that are limited only by human imagination and creativity. This lecture will focus on the visualization and therefor on the graphic representation of Data. The use of visualization to present information is not a new phenomenon. It has been used in maps, scientific drawings, and data plots for over a thousand years. E.g. Mapping the 1854 London Cholera Outbreak: Dr. John Snow. By analyzing existing data visualisation trends we will look for data visualization patterns and look at graphic design skills that are necessary to understand and produce aesthetic data visualizations.

Prof. Dr.-Ing. Joachim Charzinski: AI in Mobile Communication

Mobile Communication Networks are large and complex systems. Since the 1990s, network operators have been using AI to improve the efficiency (effect per effort) of managing their networks. The lecture will give an overview of the different areas in which AI methods are used to support non-realtime management tasks as well as real-time scheduling tasks in mobile networks.

Prof. Dr. Jens Hahn: GPU Accelerated Game AI

AI in games nowadays involves much more than simply moving pieces on a chessboard or finding the shortest path through an arbitrary maze. In modern game AI, a plethora of different methods are employed, such as path planning, decision making, strategic thinking, artificial neural networks, machine learning, and procedural content generation. This is not surprising, since the ever increasing fidelity of modern games have also elevated the requirements for convincing and believable AI to never before seen heights. These trends are not just limited to software either, since, for example, the latest generation of graphics cards features turing tonsor cores, specifically tailored towards deep learning applications. The lecture focuses on a handful of recent developments and examples in the field of Game AI and aims to inspire further investigation and experimentation

Descriptions of the other lectures will be available soon